A fluorescent region having greater brightness is displayed without a change even if the observation distance is changed, thus enhancing the quantitativeness of observation. Provided is an image processing device (6) including an illuminating portion (4) that irradiates a subject (A) with illumination light and excitation light; a fluorescence image-acquisition portion (5) that acquires a fluorescence image (G 1 ) by capturing fluorescence generated at the subject (A); a return-light image-acquisition portion (5) that acquires a return-light image (G 2 ) by capturing return light returning from the subject (A); a color-image generating portion (18) that generates a plurality of color images by adding different types of color information that constitute a color space to the acquired fluorescence image (G 1 ) and return-light image (G 2 ); and an image combining portion (20) that combines the plurality of color images that have been generated, wherein at least one of the fluorescence image (G 1 ) and the return-light image (G 2 ) is subjected to, by the color-image generating portion (18), correction processing in which exponents for distance characteristics, which are approximated to exponential functions, for the fluorescence image (G 1 ) and the return-light image (G 2 ) are matched with each other.